PROXY EVIDENCE FOR THE ROLE OF GLOBAL TELECONNECTIONS IN MIDDLE EAST HYDROCLIMATE VARIABILITY

Transcription

1 PROXY EVIDENCE FOR THE ROLE OF GLOBAL TELECONNECTIONS IN MIDDLE EAST HYDROCLIMATE VARIABILITY Yochanan Kushnir, Lamont Doherty Earth Observatory Mordchai Stein, Geological Survey of Israel! Paper published in QSR (2010) GSA ANNUAL MEETING, OCTOBER

2 The Dead (Salt) Sea! Paleo-indicator of hydroclimate variability in the Levant Source: Bookman et al. (2004) 2

3 Dead Sea Levels LGM to present Dead Sea level (DSL) curve is a composite based on geological evidence from paleoshore levels and lake sediment cores. The record was constructed in studies by Bartov et al. (2002), Migowski et al. (2006), and Bookman et al. (2004). scale discontinuity Younger Dryas Bölling-Alleröd warm period LGM Maximum 26-24ka 3

4 The backdrop: The Mediterranean - a hotspot of projected Subtropical Drying 4

5 Mt. Hermon Jerusalem Rainfall P and E climatology: rainfall (positive values) and potential evaporation (negative values) in mm Jerusalem wet 400 mm Dead Sea Watershed Lake Lisan Jericho Dashed line = approx. Ein Gedi Dead Sea boundary of Ze elim dry the Levant Jerusalem dry Amman Mean = 607 mm Source: Overview of Middle East Water Resources, USGS,

6 The instrumental record: Central Levant multidecadal rainfall variability Low-pass, ranked (values between -0.5 and 0.5), October to April rainfall variations in different Levant rain gauge stations during the 19th and 20th century.! Brown colors indicate negative anomalies (dry years) and green ones are positive (wet years). 6

7 The instrumental record: Central Levant multidecadal rainfall variability Low-pass, ranked (values between -0.5 and 0.5), October to April rainfall variations in different Levant rain gauge stations during the 19th and 20th century.! Brown colors indicate negative anomalies (dry years) and green ones are positive (wet years). dry wet dry wet d r y The (relatively short) instrumental record displays a coherent multidecadal pattern in Levant rainfall variations. 6

8 Hemispheric rainfall teleconnections Latitude 0 30 N 60 N Atlantic Ocean W 90 W 60 W 30 W 0 30 E 60 E 90 E Longitude correlation Annual (Oct-Sep) Jerusalem precipitation correlated with precipitation elsewhere. Time series were smoothed by 1 pass of a 2-nd order binomial filter. Precipitation from GPCC Areas with significant correlations are circled. What is causing this hemispheric pattern 7

9 Jerusalem rainfall -- Atlantic SST link Jerusalem ppt: annual anomaly(bars) and its 20-yr lowpass series (red line) and lowpass North Atlantic SST anomaly.! Correlation with SST (right) and SLP (below) (Data: Oct Sep 1857 to Oct Sep 1997)! SLP Anomalies are wrt the mean SST! 8

10 Atlantic Multidecadal Variability (AMV) Ting et al. (2009) Top: Observed, annual mean surface air temperature anomaly ( C) regressed on the 20th century AMV index. Bottom: The same as above but for precipitation (mm/mo).! AMV links winter (Mediterranean) and summer (African summer monsoon) variations over land because its SST footprint varies little throughout the year. ITCZ shifts northward 9

11 Cyprus Low: The raincausing mechanism cold air flow 500 hpa gph (m) cyclogenesis Right (top & bottom): The short-term (synoptic) pattern. Composite maps for 10 days with heavy rainfall in Israel Below: the seasonal patter is determined by the underlying synoptic events 60N NOAA-CIRES/Climate Diagnostics Center SLP (hpa) Ziv et al. (2006) cold air flow cyclogenesis 57N 54N 51N 48N 45N 42N 39N 36N 33N 30N 27N High pressure 60 more 40 cold air aloft 20 0 more cyclones 10W 5W 0 5E 10E 15E 20E 25E 30E 35E 40E 45E 50E 500 hpa difference between rainy and dry winters in Israel (c) -40 Low pressure 10

12 AMV winter circulation anomalies eq hpa gph (m) eq SLP (hpa) Regression of the wintertime 500 hpa and sea level pressure fields on the AMV index When the AMV is in its warm phase both surface and upper level wintertime anomalous flows over the Eastern Mediterranean (EM) are directed from south to north indicating a weakening of cold-air supply from the north and the potential for cyclogenesis is reduced. In both fields there is anomalous low pressure over the western Mediterranean and anomalous high pressure over the EM. 11

13 Holocene Dead Sea Levels SILL Litt et al. (2012) 12

14 Epochal Levant-Sahel-Atlantic link: Insolation-driven changes Radiation E.Atlantic N.African lakes ODP Core 658C Sahara-Sahel Dead Sea Lake Level July 65 N & N. Atl. SST ) (meters below msl) (Wm -2 ) Insolation 65 N, July Age cal. years BP alkenone PC sst (deg C) Kim et al. (2007) EOF alkenone (62%) High Intermediat Low sill level High 80 N 60 N (1 ) (2 ) (3 ) (4 ) Liu et al. (2007) Bookman et al. (2007) 40 N 20 N (5 ) (6 ) (7 ) (8 ) (9 ) W 60 W 0 60 E Correlation of Alkenon SST with PC1 13

15 Simulation of precession forcing - winter (2005) Between 7ka BP and the present, as North Atlantic SST cooled, SLP rose over the E subtropics intensifying the mean northerly flow of wintertime cold air from Europe into the Mediterranean Basin. 14

16 Simulation of precession forcing - summer 6K 0K change in lake mass Left: Simulated 6K 0K change in lake mass, expressed as a percentage of 0K lake mass such that a 100% increase indicates a doubling of lake mass (Tierney et al. 2011).!! Below: Present-day correlation.! Latitude 0 30 N 60 N W 90 W 60 W 30 W 0 30 E 60 E 90 E Longitude 15

17 Millennial variability B-A, YD & several other millennial lake-level fluctuations exhibit anti-phase behavior (colors indicate wet/ dry epochs in the Levant) F. Gasse (2005) However, some rapid drying events are in phase DSL m bmsl higher resolution Bookman et al. (2007) 16

18 Millennial variability B-A, YD & several other millennial lake-level fluctuations exhibit anti-phase behavior (colors indicate wet/ dry epochs in the Levant) F. Gasse (2005) However, some rapid drying events are in phase DSL m bmsl higher resolution Bookman et al. (2007) 16

19 Millennial variability B-A, YD & several other millennial lake-level fluctuations exhibit anti-phase behavior (colors indicate wet/ dry epochs in the Levant) F. Gasse (2005) However, some rapid drying events are in phase DSL m bmsl higher resolution Bookman et al. (2007) 16

20 Millennial variability B-A, YD & several other millennial lake-level fluctuations exhibit anti-phase behavior (colors indicate wet/ dry epochs in the Levant) F. Gasse (2005) However, some rapid drying events are in phase DSL m bmsl higher resolution Bookman et al. (2007) 16

21 Late Holocene fluctuations Enzel et al., 2003 Late Holocene Dead Sea Level fluctuations (Enzel et al., 2003) are correlated with Alpine glacier advance/retreat record (Holzhauser et al., 2005) - a proxy for Atlantic SST fluctuations (Denton and Broecker, 2008) and with Keigwin (1996) Sargasso SST n extens'ion Holzhauser et al., 2005 gel C 0, 'a /27 1o00.~t CO I BC BronzeAgej Age AgNRmnAe MiddlAgo AD Modem Age 17

22 Late Holocene fluctuations Enzel et al., 2003 Late Holocene Dead Sea Level fluctuations (Enzel et al., 2003) are correlated with Alpine glacier advance/retreat record (Holzhauser et al., 2005) - a proxy for Atlantic SST fluctuations (Denton and Broecker, 2008) and with Keigwin (1996) Sargasso SST n extens'ion Holzhauser et al., 2005 gel C 0, 'a /27 1o00.~t CO I BC BronzeAgej Age AgNRmnAe MiddlAgo AD Modem Age 17

23 2- higher Late Holocene fluctuations lake level Enzel et al., 2003 to sea surface changes during the past several decades, it is Relative reasonable to interpret the foraminiferal isotopic data mostly in terms of SST change. changes 18O in lake level data from each subcore are plotted together on a calendar time scale (27), it is clear When the that the same features are present in each subcore (28). Within each subcore 18O values reach a minimum ~500, 900, and 1100 years ago (Fig. 4A). Using these data, I solved the paleotemperature equation (26) after applying Deuser's disequilibrium correction (18) of +0.2 per mil to the 18O -2 mil. I then stacked the value of G. ruber and assuming that the average salinity was 36.5 per temperature proxy data from the two subcores by averaging results in 50-year bins (Fig. 4B). In general, these results indicate that there have been century-scale changes in SST of 1 to 2 C throughout the past few thousand years in the Sargasso Sea. In the-3last- half of the record there was a r 1.5 C oscillation from a minimum SST 1500 to 1700 years ago to a maximum 900 to 1000 years ago, to a minimum 300 to 400 years ago. Since the Little Ice Age, SSTs in the northern Sargasso Sea increased by ~1 C. Actual SST changes may have been even greater than indicated in Fig. 4B, in that the sediment may have been mixed differentially by burrowing (~5 cm) as sedimentation rates changed, and because stacking the 18O data may have attenuated the signal. From the raw 18O data of BC-004D (Fig. 4A), calculated SST 350 years ago was 21.5 C, about 1.5 C colder than the modern annual average. Late Holocene Dead Sea Level fluctuations (Enzel et al., 2003) are correlated with Alpine glacier advance/retreat record (Holzhauser et al., 2005) - a proxy for Atlantic SST fluctuations (Denton and Broecker, 2008) and with Keigwin (1996) Sargasso SST Y. Enzel et al. / Quaternary Research 60 (2003) lower lake level Fig. 4. (A) Oxygen isotope ratios of the n surface dwelling planktonic foraminifera G. ruber fromholzhauser Bermuda Rise BC-004 plotted et al., 2005 versus calendar age. Open symbols, BC-004A; solid symbols, BC-004D. Bars above the abcissa are a schematic C representation of proxy data for episodes of 0, glacial expansion in southern Norway (35) 1926/27 'a and summer temperature variability in 0 Fennoscandia reconstructed from tree rings (34). For the tree ring data, temperature CO maxima and minima are shown by thicker 1970 bars. For the glacier data, the bars represent 1982 two early periods of expansion (but not to LIA limits), followed by the range of age estimates for attainment of the LIA 2002 maximum (35). These terrestrial data, which are downstream of the North Atlantic, are generally consistent with the 18O data (maxima 1500 correspond to1000 cooling; minima to BC 500 warming). (B) Sea surface temperatures 18O data Age AgNRmnAe calculated from the BronzeAgej in (A), after 2000 AD gel extens'ion Keigwin, o00.~t.2000 I MiddlAgo 1500 Modem 2000 AD Age 17

24 Summary Instrumental observations and paleoclimate proxies reveal a longterm, naturally occurring pattern of hydroclimatic variability in the Levant in anti-phase with changes in North Africa (Sahel), and inphase with Europe (Alps) and North America. This pattern of variability is orchestrated by multidecadal to millennial SST changes in the North Atlantic No. Atl. SST (-) = Levant ppt (+) = Sahel ppt (-) = Alps snow (+) = No. Amer. ppt (+) Atlantic SST impact is carried over to the Levant by an atmospheric bridge : enhanced high pressure over the East Atlantic when SSTs are cold, forces cold air bearing troughs to deepen over the east Mediterranean, enhancing regional cyclogenesis and bringing more rain to the Levant. There are abrupt events (which are associated with millennial icerafting events in the No. Atl.) that break the anti-phase Levant-Sahel pattern. 18